brianhansonxyz/clearpilot
1
0
Fork 0
Code Issues Pull Requests Actions 4 Packages Projects Releases Wiki Activity

perf: pin dashcamd to cores 0-3 (little cluster)

Browse Source 
Previously ran unpinned (affinity mask 0xff) across all 8 cores. When it
landed on core 4 (controlsd) or 5 (plannerd/radard) or 7 (modeld), its
70MB/s frame copies and MP4 muxing caused cache/memory-bandwidth
contention with the RT-pinned processes. SCHED_FIFO prevented direct
preemption but not the cache thrash.

OMX offloads actual H.264 work to hardware so the main thread is
lightweight — fine on the little cluster.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
This commit is contained in:
Brian Hanson
2026-04-16 22:35:02 -05:00
parent 2ddb7fc764
commit 6dede984dc
2 changed files with 16 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

BIN
selfdrive/clearpilot/dashcamd
View File

Binary file not shown.

16
selfdrive/clearpilot/dashcamd.cc
View File

@@ -38,7 +38,9 @@
#include <cstdio>
#include <ctime>
#include <string>
#include <errno.h>
#include <signal.h>
#include <sched.h>
#include <execinfo.h>
#include <sys/stat.h>
#include <sys/resource.h>
@@ -121,6 +123,20 @@ int main(int argc, char *argv[]) {
signal(SIGABRT, crash_handler);
setpriority(PRIO_PROCESS, 0, -10);
// CLEARPILOT: pin to cores 0-3 (little cluster). Avoids cache/memory-bandwidth
// contention with the RT-pinned processes on the big cluster:
// core 4 = controlsd, core 5 = plannerd/radard, core 7 = modeld.
// OMX offloads actual H.264 work to hardware, so the main thread just copies
// frames and muxes MP4 — fine on the little cluster.
cpu_set_t mask;
CPU_ZERO(&mask);
for (int i = 0; i < 4; i++) CPU_SET(i, &mask);
if (sched_setaffinity(0, sizeof(mask), &mask) != 0) {
LOGW("dashcamd: sched_setaffinity failed (%d), continuing unpinned", errno);
} else {
LOGW("dashcamd: pinned to cores 0-3");
}
// Ensure base output directory exists
mkdir(VIDEOS_BASE.c_str(), 0755);